RoboMove: crowd-trained humanoid robot policies logged onchain
Quote294 words42 likes
Explains RoboMove, where users tweet commands to a livestreamed Unitree G1 and attempt videos feed a learning loop that updates the robot policy. Every command and action is logged onchain, creating provenance and a path to a marketplace for robot policies.
Key claims
CODEC reached a $25M mcap after the RoboMove announcement
Robot is a Unitree G1 humanoid
Training process is recorded onchain for provenance
Up another 42% and the robotics meta is just getting started.
$CODEC at $25M mcap now after the official announcement around @RoboMove, an exciting demo that turns Twitter into a public training ground for robots.
Built on @codecopenflow's dev stack, RoboMove allows the public to control and train live-streamed robots through Twitter.
The underlying core mechanism is a continuous learning loop where users tweet commands, the robot attempts them, and videos of the attempts are shared as replies on Twitter, allowing the robot (a Unitree G1 humanoid) to learn from crowd-sourced feedback, and with an outcome that ultimately updates the robot's core AI "policy."
Every tweeted command and resulting robot action (the entire training process) is recorded on-chain, creating an immutable, verifiable log of the entire training process, the model's "provenance."
This provenance is what matters, turning a black-box AI model into a transparent, auditable asset that can be owned, trusted or insured.
The end goal is a permissionless marketplace for these robot "policies", creating a novel, open economy for autonomous operators with publicly verifiable, AI-driven skills, and allowing users to buy/sell proven robot behaviours, license specific task modules, or create new challenges for rewards.
If you're still not paying attention idk what to say at this point anon. But if you like the experiment outlined above, make sure to stay tuned. Bc what starts with a single humanoid robot will see more robot types coming, which will all learn in the open with the community.
Study robotics.
Study codec.
Official RoboMove announcement linked below.
If you're yet to DYOR on the robotics narrative, check the resources below, incl. this thread by @0xPrismatic.
Or this thread here (also linking to a robotics deep dive by @cot_research in the end).
"The Robotics Trade" by @Cryptotrissy.
QUOTING @zerokn0wledge_
Do you now understand what I mean when I say $CODEC is coded anon? We're up another 78% and still early. VLAs play a crucial role to enable: - Desktop-based automation with AI operators - Robotics (robots that can see, perceive, reason & act) - In-game AI agents & autonomous g
Do you now understand what I mean when I say $CODEC is coded anon? We're up another 78% and still early. VLAs play a crucial role to enable: - Desktop-based automation with AI operators - Robotics (robots that can see, perceive, reason & act) - In-game AI agents & autonomous g
1 image on X